--- In [email protected], Michael 
Poulin <[EMAIL PROTECTED]> wrote:
>
> Please, define "Practical SO solutions" vs. unpractical solutions.

Practical vs. theoretically pure. Accepting that not everything will 
perfectly fit the guiding principles. Pure SO doesn't allow an 
unwrapped legacy application to participate in a system solution. 
Practical SO allows that a legacy application can be a useful 
component, even if might require more management attention (just an 
example).

The wrapping of a legacy application, to "service enable" it, is an 
integration exercise--essentially creating an adapter.

Another example. Data relationships are often taken to 3NF (or 
further) when modelling. A pure approach would implement the 3NF. A 
practical approach will violate 3NF for operational efficiency 
purposes. Such an approach requires more diligence so that the model 
doesn't fall into a mess.

> Iwould not call service aggregation an integration of services. 
> Service do NOT NEED to be integrated. 

This gets into defintions I suppose. Strictly speaking, two services 
that interact are "integrated."

An ERP application is often referred to as "an integrated 
applications" consisting of modules and components that are 
integrated.

This isn't to say I have an extreme view that one method of a class 
calling a method of another method within an application means they 
are integrated. *Really strictly* speaking, they are, but noone 
thinks of it that way.

SO is a way to formalize and make more consistent the approach to 
integration of capabilities. "Separately standing interfaces" is an 
integration principle, IMO. Services are designed and constructed 
with integration in mind *up front* instead of as an after thought.

-Rob


Reply via email to